Franka · Pick¶
Agent runs
Each mode of this task runs once per run. Est. cost is tokens at list price (data/prices.yml), never a bill; see the MolmoSpaces runs for every task of a run.
Codex CLI 0.157–0.159 + GPT-6 Luna, reasoning medium (OpenRouter)default runclosed
| Trial | Started | Agent time | Model requests | Tokens in / out | Est. cost | Grade |
|---|---|---|---|---|---|---|
| U✓ 17m | 10-03 21:57 | 17m | — | 3.9M / 35k | $0.066 | deterministic 1, grader_error 0, n_actions 238, replay_success 1, video_rendered 1 |
| L✗ 10m | 10-03 20:28 | 10m | — | 3.1M / 47k | $0.062 | deterministic 1, grader_error 0, live_success 0, n_actions 303, replay_success 0, video_rendered 1 |
Task instruction (upstream)
Pick up the wooden pestle with jagged top.
What MolmoSpaces states about this task
| Success criteria | 1. the object touches the robot and nothing else, and it is at least 1 cm higher than where it started 2. judged at the end of the episode: the trajectory's last row (privileged), done (standard) or 303 control steps (the benchmark's 20 s horizon), whichever comes first3. unlimited (privileged): both fresh-process replays of the handed-in trajectory end in the same state, and the check holds on it 4. limited (standard): the one episode (no reset) is recorded by the service and replays to the same state; the check holds on it live and in the replay |
| Family | pick |
| Robot | Franka FR3 with a Robotiq 2F-85 gripper on a fixed base (the DROID setup) |
| Category | Franka |
| Instance | molmospaces-bench-v2/20260415, package procthor-objaverse/FrankaPickHardBench/FrankaPickHardBench_20260206_json_benchmark, episode 12 |
| Deliverable | /app/output/trajectory.npz with actions: float64 (T, 8), one row per control step, the targets of upstream's joint-position controllers |
| Reference Solution | None is shipped. Upstream's scripted experts (molmo_spaces/policy/solvers) are reference solutions and are not in the image; neither are grasp files nor the public MolmoBot trajectories (agent egress: the model APIs only). |
| Limited Mode | Standard-mode twin of molmospaces-pick-i00-privileged (the same frozen MolmoSpaces episode): Pick up the wooden pestle with jagged top. The agent gets only the eai-standard/2.2 client (docs/STANDARD_MODE_2_2.md); the simulator runs in the sim sidecar (environment/docker-compose.yaml), which owns the episode, serves cameras, proprioception and upstream's kinematic model, and records every executed row. The collect hook (environment/sim/finalize.sh) ends the episode, lets the service exit, replays the trajectory in two fresh processes and writes final.json; the verifier grades those artifacts in a separate sandbox (tests/Dockerfile). |
| Oracle | none — no reference solution (MolmoSpaces' planners and grasp files are not shipped); positive example graded 1 by the separate verifier: molmospaces-pick-i02-privileged__wN67cFo (run codex-gpt6_luna-medium, batch molmospaces-luna-1003; https://embodied-agent-interface-v2-internal.github.io/runs/molmospaces/codex-gpt6_luna-medium-openrouter/pick/unlimited/); human review in PR #47 |
| Base Image | ghcr.io/mll-lab-nu/eai-molmospaces:0.2.0 |
| Agent Budget | 3600 s of wall clock per mode |
| Task Dirs | molmospaces-pick-i00-privileged, molmospaces-pick-i00-standard |
From https://github.com/allenai/molmospaces @ molmo-spaces 0.2.9 (benchmark molmospaces-bench-v2/20260415), as defined in our task definitions @ 030f55607.
Tags¶
Why this task is interesting¶
A wooden pestle with a jagged head stands on a counter close to a wall, among other objects. Finding it is easy; reaching it is not: a top-down grasp catches on the head and the wall leaves little room for the wrist, so the approach has to be planned.
Capability notes¶
Not yet written.
Oracle demo review¶
No demo.
Discussion¶
The hardest of the three pick candidates: an agent that sees the pestle clearly still has to find an approach the wall allows. (@williamzhangNU)